Home > mythbuster > SolidWorks Mythbuster: fully defined sketches and rebuild speed

SolidWorks Mythbuster: fully defined sketches and rebuild speed

November 18th, 2009 Leave a comment Go to comments

Have you ever heard anyone say (in the well-intentioned interest of best practice) that fully defined sketches solve faster than underdefined sketches? I have. And I know these people only meant good by it, but the statement had the smell of something fishy. When things just don’t quite add up intuitively, that’s a good time for a test. And that’s what Chapter 8 of the SW bible does.

I made a sketch that was fully defined bunch of rectangles. And then I copied it, and removed the relations and dimensions. And then I looked at the Feature Statistics:

sketchrebuild

Ok, so it’s still a good idea to fully define sketches, but don’t do it in the name of rebuild speed. Also, you need to keep a little perspective on things. Rebuild speed isn’t always the utmost goal to achieve. There are some things you can’t do without, and sketch relations and dimensions are two of them.

Categories: mythbuster Tags:
  1. November 18th, 2009 at 15:50 | #1

    great myth buster matt! thanks for sharing!

  2. November 18th, 2009 at 17:59 | #2

    Matt-

    Very informative. A very few of my clients require fully defined sketches.

    Devon

  3. Steve Calvert
    November 19th, 2009 at 10:35 | #3

    Matt, you need some kind of watermark that says “Busted” so you can show it behind the myths you’ve busted.

    Good job..

    Steve

  4. November 19th, 2009 at 11:34 | #4

    @Steve Calvert
    Your wish is my command. Great idea.

  5. Jeff Cox
    November 19th, 2009 at 11:44 | #5

    Point taken, but undefined sketches still make me cringe!

  6. November 19th, 2009 at 14:09 | #6

    Although it may not help rebuild speed I still consider it “mandatory where possible”. This is because I have seen many sketches blow up because they were undefined even if a feature totally unrelated to the sketch is what is changing. Whether it was a radius flipping inside out or a spline going nuts it happens and I preach to all of our designers and engineers to fully define sketches whenever humanly possible! :)

  7. Stephen Burke
    November 19th, 2009 at 14:33 | #7

    Would this be the same for mates? (Perhaps a question for another mythbusters post)
    AFAIK fixed mates rebuild fastest, and limit distance mates are the slowest.

    Excellent idea about this mythbusters section.

  8. November 20th, 2009 at 11:40 | #8

    How do these compare to the same sketch with completely undefined by fixed entities?

  9. November 20th, 2009 at 13:19 | #9

  10. November 20th, 2009 at 13:22 | #10

    @Stephen Burke
    I’d guess that mates are similar, but with a bigger rebuild hit, and the different types of mates would have a wider range of rebuild speeds. Maybe that’s a mythbuster that I can take on in the near future.

  11. November 20th, 2009 at 14:10 | #11

    One thing I’ve been noticing (here and elsewhere), total rebuild times are varying quite a bit for similar or same situations. I’m sure you noticed the .16 vs the .05 seconds between these two experients.

  12. November 20th, 2009 at 14:12 | #12

    @Stephen Burke The more mates, the worse your rebuild times. Don’t fully contrain screws (if you happen to want your screws in the model),

  13. November 20th, 2009 at 14:13 | #13

    @fcsuper BTW, I’ve done this test before, if Matt doesn’t mind me posting a link to that old article.

  14. Ray Regan
    November 21st, 2009 at 08:17 | #14

    No Blue Sketches! Tom Hagerty nailed it in his comment. I’ve seen it numerous times, when you modify a model with underdefines sketches, The model goes, See Ya in a Kia, Cra-A-Zee!

  15. November 21st, 2009 at 12:07 | #15

    @Ray Regan
    This is a myth that I totally want to bust so bad I can taste it! The idea that undefined sketches move at all is preposterous. Partially defined sketches I can see moving unpredictably, but a totally undefined sketch should not move at all.

    Does anyone have any ideas on how I can either bust or confirm this?

  16. November 21st, 2009 at 12:08 | #16

    @fcsuper
    Matt, rebuild times that I show are all over the map because I am often using different sketches, on different computers with different versions of the software. The only thing that really stays the same are the percentages.

  17. smartin
    November 23rd, 2009 at 08:56 | #17

    @matt

    I think you and Ray are talking the same thing when you say ‘partially defined’ and he says ‘underdefined’ (not undefined). I’ve seen underdefined used in this context before, and while it’s not perhaps as clear as saying ‘partially defined’ I’ve always taken it to mean the same.

    My guess is that when users complain about ‘completely undefined’ sketches moving they are not noticing that the sketch picked up some relations to other sketches/geometry in the model.

  18. November 23rd, 2009 at 10:16 | #18

    Undefined sketches are good with Instant3d. They do jump around when trying to manipulate sketch entities via drag methods.

  19. November 23rd, 2009 at 12:55 | #19

    @matt

    Matt- I agree with you. Another myth to bust, the one that says Parts with no Mates move by themselves in SW Assembies.

    Great post.

    Devon

  20. November 23rd, 2009 at 16:17 | #20

    I’ve seen this myth over and over. Perhaps it’s a bit more true when a whole feature tree relying upon these undefined sketches must be solved—not in a single, non-tree-related sketch is undefined? Not sure, but I think that would be the ultimate test. How fast does a series of extrusions—built upon undefined sketches—rebuild vs. the fully-defined sort?

    My guess is that anyone saying sketches should always be fully defined isn’t doing a lot of work with splines (such as surfaces-heavy modeling), and certainly not doing lots of wild editing of swoopy geometry. I typically don’t define splines with anything more than relations. I often go into the sketch and push/pull spline points to tweak the overall form of a surface fill/loft/sweep/whatever.

    Further, this has never caused problems of things moving around. Sure, if you’ve got a point of a sketch segment pegged to exterior geometry (and little else defined), when you move that external geometry, part of your undefined sketch will also move. But isn’t this also predictable (and even sometimes desired)?

  21. November 25th, 2009 at 14:11 | #21

    I work with splines all of the time and also try to constrain them completely as well. If I need to change it by dragging I may delete some of the constraints to push and pull it but once it is set I lock it down again. I have seen the result of this ending up in a blow up model way too many times for me to not try and avoid blue sketches. Like I said, whenever possible, so it is not a steadfast rule but one I probably stick to 99% of the time.

  22. December 3rd, 2009 at 14:47 | #22

    I would agree with Tom Hagerty and Ray Regan. Blue sketches and black sketches are there for a reason…

    Under defined sketch/es can cost you 10′s if not 100′s of thousands of dollars never mind the loss of time etc. During crunch time when your brain is almost fried and drawing are due to be released for tooling… All you need to do is to slightly move a line or a point in one of your under defined sketches. How would you even know what you did or moved and even if did find what moved how would you place it back where it was? Ctrl Z doezn’t always work the way you want…

    Blue sketch to me is like a red cloth to a bull. When I teach, this is number one rule I demand my students to learn. NO BLUE SKETCHES!!!

    Perhaps, I would agree on blue sketches during conceptual stage when nothing is critical, but for production……no way, keep em black I say.

  23. Chris K
    December 10th, 2009 at 12:50 | #23

    Fully agree with Alex. Is SW really rewarding bad practices with lower rebuild times? Improving stability and reliability of the model by proper modeling should not be a compromise with rebuild time (or anything else for that matter).

    Devin: about the myth that part move by themselves: parts don´move until your manager (or your idiot colleague) grabs the mouse and tries to rotate the view. It is all about reliability. How sure are you the model is correct when you´re about to spend 10 times your yearly salary on moulds? Believe me, you want to fully mate parts and fully define sketches.

    Jeff: seems odd to me that there is no simple way to fully define splines AND be able to manipulate them as though it were clay. Another myth to bust: the one that says that SW is user friendly.

  24. John H
    March 12th, 2010 at 06:00 | #24

    I’m not sure the test methods used prove the point entirely regarding whether sketch rebuild times are better or worse if fully/underdefined.

    It would seem obvious to me that a sketch with no relations will solve faster than a fully defined one, because there is nothing to solve! However, there may be an intermediate case with partially defined complex sketches, where the solver might struggle to reach a solution and therefore take longer.

    In general I would go along with the others that say fully define sketches. There is nothing more time-wasting than an unpredictable outcome!
    Actually, that’s one of my major gripes with Solidworks, that fully constrained assemblies don’t always behave predictably – especially multi-level ones with different sub-assy configs, flexible sub-assys etc.

  1. No trackbacks yet.

Upload Files

You can include images or files in your comment by selecting them below. Once you select a file, it will be uploaded and a link to it added to your comment. You can upload as many images or files as you like and they will all be added to your comment.

29 visitors online now
19 guests, 10 bots, 0 members
Max visitors today: 36 at 09:16 am EDT
This month: 48 at 09-02-2010 01:16 pm EDT
This year: 64 at 05-16-2010 10:32 pm EDT
All time: 64 at 05-16-2010 10:32 pm EDT